AI Jailbreak & Prompt-Injection Red Teamer (LLM Security Specialist)

About the role

We're looking for a skilled AI security specialist to join our red teaming efforts and help stress-test large language model (LLM) systems before they reach production. You'll act as an adversarial researcher, probing AI models and their surrounding tool ecosystems for jailbreaks, prompt injection vulnerabilities, and tool-use abuse patterns. Your findings will directly inform safety guardrails, alignment training data, and defensive engineering priorities.

This is a remote, global engagement ideal for someone who thinks like an attacker but works like a researcher — methodical, well-documented, and focused on making AI systems genuinely safer.

What you'll do

• Design and execute ethical jailbreak attempts against production and pre-release LLMs to uncover safety bypasses
• Craft and test prompt injection attacks across single-turn, multi-turn, and agentic/tool-use contexts
• Identify and document tool-use abuse vectors, including unauthorized function calls, data exfiltration paths, and permission escalation
• Build reusable red-teaming playbooks, taxonomies, and severity-rated vulnerability reports
• Collaborate with model safety and alignment teams to translate exploit findings into mitigations, guardrails, and training signal
• Track emerging jailbreak techniques and adversarial prompting trends across the LLM security community
• Clearly communicate technical findings to both technical and non-technical stakeholders

Requirements

• Demonstrated hands-on experience with LLM red teaming, adversarial prompting, or AI safety evaluation
• Practical knowledge of prompt injection techniques and jailbreak methodologies
• Understanding of tool-use/agentic AI architectures (function calling, plugins, RAG pipelines) and their abuse surfaces
• Strong written communication skills for producing clear, structured vulnerability reports
• Ability to work independently in a remote, asynchronous environment
• Familiarity with responsible disclosure and ethical hacking principles

**Nice to have:**

• Background in cybersecurity, penetration testing, or AI safety research
• Experience with major LLM providers' APIs and safety filters
• Contributions to public jailbreak/red-teaming research, CTFs, or bug bounty programs

Compensation

Hourly rate: **$50–$90/hour**, commensurate with demonstrated expertise. Flexible, remote engagement open to candidates worldwide.